Classifying drugs by their ease of absorption
Simulating gastric absorption
New theories of membrane transport
How to pick the winners
Current thoughts and research in improving drug absorption provided the
theme of the one-day biopharmaceutics symposium organised and chaired by Dr
Rosemary Leak (Glaxo Wellcome, Ware) at the Conference on September 12
Classifying drugs by their ease of absorption
The most common and simple route of administering drugs is by oral ingestion.
To be efficacious by this route the drug needs to be optimally absorbed; that
is, the preferred chemical form should be transported rapidly and completely
across the gut wall, without metabolism or degradation. Consequently, it has
long been the goal of researchers in drug development to design new chemical
entities that are both pharmacologically active and have biopharmaceutical properties
that favour drug absorption.
Professor Bill Charman (Monash university, Victoria, Australia) opened the session
with a comprehensive review of a proposed classification of the oral biopharmaceutical
properties of drugs.
In the early 1990s, researchers into factors affecting oral absorption began
to evolve a scheme for classifying drugs according to the likelihood of their
being absorbed. Professor Charman noted that such a classification would be
helpful for regulatory authorities faced with deciding the extent to which additional
testing of generic products would be necessary. A simple system proposed by
Amidon and others in 1994, classified drugs as having high or low solubility
and high or low permeability. Professor Charman said that, although other factors
had been proposed for inclusion in this classification scheme, he believed the
original proposal (with suitable refinement) served remarkably well. Draft guidelines
issued by the Food and Drug Administration had been finalised in the past two
weeks and the robustness and usefulness of the final version was a tribute to
the careful thought that had been put in by the advisers.
The classification system placed drugs into four classes (see Figure 1), although
Professor Charman pointed out that, as with geographical boundaries, the edges
of these classes could not be considered rigid. Class I drugs had ideal properties
for absorption and did not usually pose a problem when demonstrating bioequivalence
of generic drugs. However, drugs that appeared in class I were not necessarily
successful drugs, as other factors that affected their development were often
more important. Nevertheless, from a biopharmaceutical standpoint, any drugs
that had biopharmaceutical properties in the extremes of class I were worth
pursuing.
The rate of in vivo dissolution was the rate-limiting step for oral absorption
of class II drugs. Hence, for such drugs, demonstration of bioequivalence of
different formulations or generics was important. In vivo/in vitro correlations
could be particularly useful for this class.
Turning to Class III drugs, Professor Charman noted that, for these, the rate-limiting
step was permeability. Consequently, rapid dissolution was necessary to maximise
the time that high concentrations of the drug were in contact with absorbing
membranes. Appropriate formulation, therefore, was critical.
Class IV drugs had significant and inherent limitations to their efficacy by
oral delivery. In short, Professor Charman had only one piece of advice regarding
class IV drugs — get another drug.
Looking to the future, Professor Charman suggested that efflux or the metabolic
properties of putative drugs could be added to the classification scheme. Coupled
with methods of rapid assessment of solubility, permeability and metabolism,
the classification would be invaluable in optimising evaluation of compounds
in drug discovery. (Top)
Simulating gastric absorption (Top)
Professor Jennifer B. Dressman (Johann Wolfgang Goethe university Institute
for Pharmaceutical Technology, Frankfurt, Germany) returned to the use of in
vivo dissolution methods to determine biopharmaceutical properties.
An important starting point was by refering to the biopharmaceutical classification
system. To predict absorption it was necessary to consider the events in the
gastrointestinal tract that had an impact on the process and which of those
then became part of any rate-limiting process.
Class I drugs dissolved rapidly and quickly permeated the gut wall. Hence,
in vitro dissolution experiments were of little help in discriminating between
drugs in this class. The only requirement for a dissolution medium for class
I drugs was that it should be wet, Professor Dressman said.
Assuming no chemical degradation took place and that there was no first-pass
metabolism, then only class II drugs could yield worthwhile discriminatory data
(ie, where a drug had limited solubility and a high permeability). The first
step, therefore, was to determine whether a drug under investigation was a class
II drug. Professor Dressman pointed out that controlled-release formulations
were often special cases of class I drugs behaving as class II drugs, and that
drugs with low water solubility administered in low doses (eg, digoxin) might
not qualify as class II drugs.
Whereas standard methods of dissolution testing were designed to discriminate
between physical forms of a drug for analytical or quality control purposes,
collecting dissolution data for prediction of biopharmaceutical properties demanded
that the dissolution medium should be as close to the in vivo situation as possible.
The more closely this could be achieved, the better the prediction for absorption
would be, Professor Dressman said.
The gastrointestinal milieu was not fixed and, for a complete simulation, a
number of different situations needed to be assessed. Professor Dressman described
her experiences with simulated gastric and intestinal fluids both for "fasted"
and for "fed" situations. A simulated gastric fluid, for example,
should contain hydrochloric acid, a surfactant and sodium chloride, made up
to a volume consistent with normal subjects with water. Sometimes consideration
of the actual volume in relation to the proposed dose level was important for
valid simulations. It was recognised that surfactants were present in gastric
juice but that their exact nature was surprisingly unclear.
To simulate the fed state, milk was added to the medium, as it contained the
appropriate protein:fat:carbohydrate ratio for the Western diet. However, it
could form complex matrices that could cause analytical problems if the analyte
had to be separated before end-point determination.
Dissolution experiments using simulated fluids had shown good correlation with
clinical data obtained from fasted and fed volunteers, providing a justification
for this approach. Professor Dressman proposed a comprehensive model for prediction
of biopharmaceutical properties into which gastric emptying and other processes
could be incorporated to predict the overall absorption profile (see Figure
2). (Top)
New theories of membrane transport (Top)
Professor Ronald T. Borchardt (department of pharmaceutical chemistry, University
of Kansas, US) emphasised the importance of the biopharmaceutical classification
of drugs.
The year 1995 had seen the old idea that absorption was a passive process being
seriously questioned. The idea of the intestinal mucosa being a physical barrier
to absorption of drugs, by-passed only by aqueous pores or lipid channels in
the membrane, had now been discarded. Modern concepts suggested that the mucosa
also provided barriers in the form of metabolic enzymes, particularly the cytochrome
P450 family, and transporting proteins such as p-glycoprotein. These prevented
permeation of drugs whose physicochemical characteristics otherwise suggested
facile transport by passive processes. Common excipients, previously considered
innocuous in the biopharmaceutical sense, might have an effect on these enzymes
and transporters and hence might affect the bioavailability of a drug in unexpected
ways. Typically, studies were carried out in vitro by noting changes in drug
concentration on either side of model membranes in diffusion cells.
The transport of drugs across membranes in the presence and absence of non-ionic
surfactants (Cremophor EL, Tween 80 and polyethylene glycol [PEG 300]) had been
studied by Professor Borchardt and his colleagues. They had found that these
excipients could affect the activity of p-glycoprotein in the intestinal mucosa,
thus affecting the protein’s ability to limit the transmucosal transport
of drugs such as taxol. For example, inclusion of 20 per cent PEG 300 in the
transport buffer typically used in Caco-2 (a cell culture model of the intestinal
mucosa) cell transport experiments totally inhibited polarised efflux of [3H]taxol.
The researchers had, more recently, focused on the mechanism of this inhibition.
Using appropriate probes, it could be shown that PEG 300 changed membrane fluidity
by its effect on either lipid side chains or polar head groups. Similar results
were found using a Madin-Darby canine kidney cell line transfected with the
gene for p-glycoprotein.
It was not yet known whether these findings were relevant to the in vivo situation
and to formulation design. However, it was now recognised by regulatory authorities
and researchers that poor biopharmaceutical properties were not acceptable,
that excipients might not be as inert as previously considered, and that excipients
could now be designed to optimise biopharmaceutical properties, for example,
by affecting P450 enzymes in the gut wall. It was significant, Professor Borchardt
said, that the Food and Drug Administration in the US had begun to ask about
the effect of excipients on the biopharmaceutical properties of new drugs. (Top)
How to pick the winners (Top)
Dr John Ayrton (Glaxo Wellcome, Ware) developed the theme of classification
and its role in allowing researchers to predict absorption properties. In silico
(ie, computer modelling), in vitro, and in vivo approaches employed by Glaxo
Wellcome were used to test permeability, although metabolism could also play
a significant part in the successful selection of candidate drugs.
In terms of biopharmaceutics, the ideal medicine was one that provided the right
balance of potency, selectivity and pharmacokinetics. Combinatorial chemistry
and subsequent high-throughput screening provided thousands of drug candidates.
The issue now was how to aid the design and selection of compounds with suitable
biopharmaceutical properties on a massive scale. The traditional drug development
process, in which a drug could fail at a late stage for pharmacokinetic reasons,
was uneconomic.
In vitro approaches, such as those already described by previous speakers, could
isolate absorption issues and speed up the selection process. In vivo methods,
because of the complexity of whole-body models for pharmacokinetic evaluation,
required increasingly complex solutions. Glaxo Wellcome had pioneered cassette
dosing. This involved giving several compounds in a mixture (typically five
to 10 components) to a single animal simultaneously. This was followed by blood
sampling and assay of the separate compounds and sometimes their metabolites.
Dr Ayrton admitted that he had initially been sceptical about this method and
the approach would not have been feasible with classical methods of analysis
(eg, spectrophotometry and chromatography). It was only by using specific and
sensitive analytical methods — mass-spectrometry, in particular —
that the cassette method could be applied. Dr Ayrton said that there might be
limitations to the interpretation of multiple dosing. For example, one component
could affect metabolising enzymes or modify blood flow and compromise the results
for other components. Nevertheless, the method did speed up the decision-making
process — which was the reason for its use — and, where an individual
compound showed promising characteristics, it could be studied separately in
more detail.
For in silico testing, the long-term aim was that chemical structure alone could
be used to predict biopharmaceutical characteristics. Once systems were in place
(and had been appropriately validated), hundreds of compounds could be run through
computer programs, predictions about compounds could be made rapidly and those
that seemed promising could be tested experimentally.
Comparisons of the various approaches to predicting biopharmaceutical properties
could be both confirmatory and complementary, emphasising the value of using
many methods in drug discovery. In the total drug research process, few compounds
now failed at a late stage because of unfavourable pharmacokinetics, Dr Ayrton
said. (Top)